The prostate is a mobile organ that demonstrates momentary and daily changes in position and shape that critically affect accurate delivery of radiation therapy. For this reason, in modern technically demanding radiation therapy for prostate cancer, it is standard practice to localize the prostate on most days treatment as an aid to achieve the goal of delivering a high dose to the prostate while protecting nearby radiosensitive normal tissues. The Phase I study established the feasibility of a novel approach that uses unique mathematical properties of a class of statistically trainable deformable shape models to estimate treatment image data by mapping reference CT image data into the treatment space based on measured positions of markers implanted in the prostate. The estimated images capture the pose, position and shape of the prostate during treatment and are suitable for accurate calculation of dose delivered to the prostate and immediately surrounding tissues. The first two years of this Phase II study will focus on design and development of a clinical prototype system based on this creative technology. The prototype will be evaluated in the clinical setting during the third year. The envisioned commercial form will accept input data from multiple devices including the treatment planning system, treatment machine, and marker tracking system, and will communicate output data to the treatment planning system. Within a few seconds the proposed prototype will use input data to automatically compute the estimated treatment image and calculate delivered dose. The proposed system will leverage work accomplished under a currently active Phase II project (R44 CA119571) that will allow the prototype to be used with intra-treatment imaging devices as well as marker-trackers. Moreover the proposed approach may enable a form of dose computation, called synchronized dynamic dose reconstruction, needed for the most accurate form of ART.